Don’t leave it too late before you get an EMC health check on a new product design.
It’s nearly time. The product launch is around the corner and the factory is ready to start churning out the orders. All you need is a pass on the electromagnetic compatibility tests required to sell the product in Europe and everything is good to go. Then the results come back: it’s failed and the alterations mean going back to the drawing board.
One way to avoid the problem is to invest in EMC pre-compliance testing to check out the product design before finding out that it needs a shielded metal casing instead of the good-looking custom-moulded plastic case that will help sell it to customers or even a revised printed circuit board (PCB) design to put in extra ground planes to stop it from becoming its own unintentional radio transmitter.
When it comes to deciding how much pre-compliance testing needs to be done, a lot depends on the nature of the product. Has it been designed and built practically from scratch or does it incorporate systems that already sport the vital CE mark?
Andy Lawson, technical supervisor at test house TÜV Süd UK, explains: “If you are building something from scratch you often have no idea how it will perform really, even with best practice. You can get problems that you had no idea could come up.”
Rob Armstrong, consultancy manager at York EMC Services, says: “If you are just updating something, where you are swapping one component for another and the behaviour of the replacement is similar then testing or a technical argument is possible. If you are assembling a new system made out of CE-marked subsystems, you still have to test. But you should be fairly confident of how those systems will behave.”
But it is worth checking, especially for conducted emissions problems. Lawson quotes the mantra of EMC work: “CE + CE does not equal CE”.
On assemblies of CE-marked subsystems, it is relatively straightforward to assess final performance and to see where problems crop up, particularly for radiation that passes down the power cables. “You can literally add one unit after another and check the emissions down the power line,” says Lawson.
More intricate assembly of subsystems can introduce further EMC issues. An example is a monitor array used for large-scale displays at airports and stadia. Often based on standard flat-panel displays, the original casings may need to be removed so the screens can be fitted together without obvious gaps. “The original CE mark for the display definitely no longer applies. So, if you are doing something like that, we would normally recommend a pre-compliance test, before full CE testing takes place,” says Armstrong.
When it comes to performing pre-compliance, the apparent choice is between doing tests at an accredited lab or trying to get an idea of how well the product will perform based on trials in the development lab itself, perhaps at night for less interference, or in a nearby basement. “People who do their own major pre-compliance work may have something quite similar to an accredited lab with an anechoic chamber and similar equipment,” Armstrong says. “If people have made the decision to have an anechoic chamber they will make it as close as they can get to an accredited lab.”
For organisations that do not work on enough projects in a year to justify the outlay on a dedicated in-house pre-compliance chamber, it is possible to buy small ready-made test chambers. A typical design is the gigahertz transverse electromagnetic chamber (GTEM). Its distinctive horn shape is intended to replicate the behaviour of a much larger chamber in terms of its ability to create a uniform electromagnetic field around the equipment under test (EUT) from an amplified signal injected at the narrow end.
“Smaller chambers can work quite well. But with something very small the interactions between the walls and EUT are quite different from what you will find with a larger chamber,” Armstrong says. “A lot depends on how good your pre-compliance chamber or test setup is. We run a health check to see how close they are to an accredited lab.”
Lawson adds: “We did have a GTEM cell for a while with the aim of renting it out to customers for use in their own labs. But everyone found that the results they got weren’t comparable enough to what they would get in a full test chamber.”
Tests that replicate the conditions of a dedicated lab are not essential, particularly when trying to identify potential trouble spots. “People will do desktop pre-compliance on things that they know that they are likely to have a problem with. If they have a plastic enclosure rather than metal, which might be for cost reasons or to make the product more attractive, they might bring a mobile phone close to it and then ring it to see what happens,” Armstrong explains.
“Obviously, it is a long way away from a full EMC test, but it can work as a simple test in terms of functionality. You can get a feeling of whether you are on the right lines. And, certainly if you are on the wrong lines, it involves less initial outlay. For checking susceptibility in a wider frequency range, people tend to swing towards more functional EMC tests: what will it be exposed to in the field? Rather than trying to replicate the EMC test conditions,” Armstrong adds.
Lawson points out that the intention behind the immunity standards is to catch problems caused by communications, so testing with a variety of wireless transmitters, ranging from DECT phones through mobiles to Wi-Fi makes sense. With radiated emissions, there is a limit to the kinds of problems that can be spotted with desktop pre-compliance. “But you can see glaring problems,” he adds.
“For testing emissions you need a quiet background, so many people do it at night. But even then you’ve got police band signals, GSM calls and air traffic communications. But you know those signals should be there. With this type of testing you are looking for unknown emissions. Then it’s easy to prove whether it’s coming from the kit or not,” Lawson explains.
In common with accredited labs, the key piece of equipment is a spectrum analyser armed with suitable antennas and probes.
Lawson says: “With the spectrum analyser, you want something that works in the 30MHz to 1GHz range: that seems to be the main trouble band. Unless you have a system operating at gigahertz frequencies, if all your problems are in that range, anything above is likely to be harmonically related to frequencies below 1GHz.”
Simplifying the checks
Companies such as Rigol, Rohde & Schwartz and Tektronix have made spectrum analysers that include functions to simplify pre-compliance. The USB-based RSA306 from Tektronix includes software that lets users put in limit lines to find unwanted peaks in electromagnetic interference (EMI) from the equipment under test. Similarly the Rigol DSA1030 includes support for limit lines as well as EMI filter and peak detector options.
Jim Martin, applications manager at Rigol’s UK distributor Telonic Instruments, says: “The instrument provides an approximation of the detector used in the standard tests; it works in the same circumstances but is not fully compliant to the standard. A fully compliant detector would be a lot more expensive.”
Probes made by companies such as TekBox provide a way to examine near-field emissions - those very close to the source components - to help track down any troublemakers in the circuit. “For susceptibility testing, you can use them to inject a radiated signal into the PCB,” Martin adds, helping to find areas of the design that may not be robust in the presence of strong RF emissions.
For susceptibility tests that rely on high signal power, the only realistic option short of investing in an anechoic chamber is to head into a dedicated lab. Systems intended for use on the rail network need to pass more rigorous testing than consumer equipment, which involves more powerful interference being inflicted on them - interference that needs to be shielded from the outside world so that it does not overcome cellular, radio and TV transmissions.
“The immunity requirements are quite high for railway equipment: higher than you would be able to generate in the office. If you haven’t got an anechoic chamber and a couple of RF power amplifiers you won’t get close to the most stringent tests needed,” Armstrong says.
Although they may not be subject to such stringent standards, motor drives are often tested to higher standards to ensure that the electrically hostile environment around large industrial motors does not cause problems.
Even computer equipment could start to be tested to higher standards to ensure it will be able to run if hit with a criminal EMI attack. However, such systems are unlikely to need the level of immunity testing to which military hardware is subjected. “The military will test at a field strength stated in a standard and they keep going until it falls over to give more of a feel, to see how much headroom they’ve got,” says Armstrong.
Military users are more likely to encounter powerful EM weapons. The power needed to disrupt a data centre from outside is significant - calling for a substantial power source and antenna. Anyone driving such a van around the City of London or Canary Wharf with plans to disrupt the financial trading networks would most likely be spotted and stopped before they got close to the target.
However, on a much smaller scale, criminals are willing to use EMI as a way of teasing out encryption keys and to bypass security measures in the electronics that they have in their hands. They may ‘fuzz’ the unit, trying to disrupt its operation using bursts of EM radiation as well as power fluctuations. Or, the case of side-channel attacks, simply try to detect pulses of radiation from the equipment as processors crunch through data. This is another area where desktop-class test equipment and near-field probes can help determine how well the equipment will fare in the hands of a would-be hacker.
Studies of the vulnerability of electronics goods have shown that careful attention to the way in which the system reacts to external interference and restricting the ability of instruments to pick up subtle EMI on the way can go a long way to ensuring the system cannot be easily reverse engineered or disrupted in operation.
Whatever the end product, pre-compliance makes sense and should not be treated as an afterthought. “What we find is that often, though certainly not all the time, EMC is forgotten until a fairly late stage in the project,” Armstrong says. “Most people do pre-compliance after the thing is finished. We try to push people to do it as early as possible. Do little pre-compliance tests during design. If you are doing it when it’s all signed off, it’s much more expensive and difficult to fix.”